HUMAN FACTORS IN THE DESIGN AND OPERATION

OF HANDWHEEL CONTROLS USED IN

A DYNAMIC MANUAL TASK

by

LARRY BERNARD JORDAN, B. E. S.

A THESIS

IN

INDUSTRIAL ENGINEERING

Submitted to tha Graduate Faculty of Texas Technological College in Partial Fulfillment of tha Requirements for

The Dagraa of

MASTER OF SCIENCE

IN

INDUSTRIAL ENGINEERING

Approved

May, 1969

hOo. S-\ rs /. TABLE OF CONTENTS

LIST OF TABLES iv

LIST OF ILLUSTRATIONS v

I. INTRODUCTION . . . • c 1

Human Factors Research Areas 3

The Force Platform 9

Purpose and Scope . • . 13

Review of Previous Research l4

II. EXPERIMENTAL PROCEDURE • . 21

Apparatus 21

Performance of the Experiment 32

III. DESIGN OP THE EXPERIMENT 39

Selection of Variables 39

Discussion of Variables Selected . « . . . 46

Design of the Experiment 52

IV. ANALYSIS OF RESULTS. . . . . . . . . . . . . . 56

Analysis of Force Traces 56

Analysis of Variance .57

Analysis of Graphic Data . .61

V. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS. . . 88

Summary 89

Recommendations for Further Research . .,. 92

LIST OP REFERENCES 9^

ii

ill

TABLE OF CONTENTS Continued

APPENDIX 97

A. Force Platform Calibration Record . . . . . 9^

B. Graph of Subject Main Effect „ 99

LIST OP TABLES

Table Page

1. Optimum Diameter of Handwheels

Various Torques and Positions l6

2. Subject Data 22

3^ Applied Equilibrium Force For Various Torques and Handwheel Sizes ^7

4• EMS for Pour-Factor Experiment in a

Randomized Block Design 5^

5. Summary of Experimental Variables 55

6. Analysis of Variance 60

iv

LIST OP ILLUSTRATIONS

Figure Page

1^ Whitney's Method 12

2, Beckman Dynograph Recorder 23

3. Force Platform 25

H, Force Platform Schematic 26

5^ Torque Device Schematic . . . . 30

6. Torque Device 31

7. Calibration Table 33

8. Torque Device and Force Platform 35

9. Experiment in Progress . 36

10, Handwheels Used 43

11, Hand Positions 45

12, Typical Dynograph Recording 58

13• Relationship Between Resultant Bodily Reaction Force and Handwheel Size 62

l4• Relationship Between Resultant Bodily Reaction Force and Resistant Torque 66

15• Relationship Between Resultant Bodily Reaction Force and Relative Height 68

l6^ Angle of Elbow Flexion for the Average Subject 70

17. Relationship Between Resultant Bodily Reaction Force and Hand Position 72

18. Relationship Between Resultant Bodily Reaction Force and Resistant Torque for Various Handwheel Sizes 76

V

CHAPTER I

INTRODUCTION

Before World War II the human operator was not

given a great deal of attention by designers of his tasks

and equipment. This is understandable when one considers

that equipment during this period was not nearly as elabor­

ate as that developed during and after World War II and

the deficiencies caused by this neglect of the human oper­

ator were not nearly as great or as noticeable. Designers

became acutely aware of the need for close coordination of

men and machines when the complexity of some types of mod­

ern military and civilian equipment began to outstrip the

abilities of the men trained to operate them. The effi­

ciency of performance that was theoretically achievable

was never realized due to the limitations of the overlook­

ed "weak link", the human operator.

The realization of the importance of the "human

factor" brought a new era in equipment design. It also

spawned a new field of scientific endeavor known as human

engineering (human factors engineering, ergonomics, bio­

technology) . This new area "is not a single scientific

discipline but a synthesis which integrates the biological

sclences--psychology, anthropology, physiology, and

madicine--with engineering (5)." Human enginaaring means

"fitting the machine to the man, and keeping him function­

ing with efficiency, with safety, and without discomfort

in any environment (l6)."

Before the advent of this interdisciplinary science,

the design of equipment was primarily the responsibility of

the engineer who was concerned mainly with engineering con­

siderations. "As often as not this did not matter vary

much because the man who operated the equipment was tha

most efficient part of the man-machine unit (9)." This

neglect of the human factor by design engineers becomes

apparent when one considers tha following definition:

Engineering design is ".,. the process of applying various

techniques and scientific principles for the purpose of

defining a device, a process, or a system in sufficient

detail to permit its physical realization (27)."

Reliable performance within an existing environ­

ment was considered a primary basis for acceptance of a

design• Also of concern to the design engineer was socio­

economic conditions.

There are long and short-time economic cycles that are related directly and indirectly to natur­al, social, business, and political factors• Par­ticular designs are formed within limited economic boundaries and associated conditions. In addition to economic influences, there are influences on designs established by various groups of people• (7)

Finally, much serious attention was directed to the

3

selection of materials based mainly upon three primary

considerations:

1. Mechanical functions of materials• 2, Non-mechanical functions of materials

such as appearance, safety, etc^ 3* Economics of material selection•

These, then, were the major factors with which design

engineers were primarily concerned in the past^ Modern

design engineers have recognized the importance of the

human factor and have added it to the list of major fac­

tors influencing design.

Fortunately, through research, a considerable

volume of valuable human factors Information has been

generated which the designer may apply to the design of

tools, equipment, and systems involving man as a user or

component. Let us now briefly examine some of the more

Important areas that have been given attention by human

factors researchers•

Human Factors Research Areas

In the early stages of human engineering it was

desired to know the various dimensions of the human which

could be used In the development of improved working con­

ditions. Research was done in the area of anthropometry

in order to determine statistical data for the basis of

man-machine relationships. The data thus obtained were

gathered together, and normal curves were developed

representing the sizes of entire individuals as well as

various components of the individuals such as arm lengths,

foot sizes, eye-level positions, leg lengths, etc., and

certain physical abilities of individuals. The human

measurements and the statistical data obtained not only

provide for a fairly good man-machine relationship, but

they also give a certain amount of insight into such

things as physical endurance, motivation, strength of

individual members of the human body, etc. Helpful in­

formation in the area of health and safety has also been

provided. (7)

The availability of this information stimulated

research in the areas of workplace design and equipment

layout. Workplaces were designed to accommodate the

varying bodily dimensions of the greatest possible per­

centage of the operator population. Researchers attempt­

ing to determine the optimum arrangement of controls and

displays found that there was a scarcity of data on the

sensory-motor capabilities of the human being. This need

sparked research into the range, strength, and speed of

human movements, studies of h"uman fatigue and endurance,

and studies of the sensory mechanisms such as vision,

hearing, etc^

It was found that, in most cases, environmental

and psychological as well as physiological or biological

factors were Involved in human performance• Consequently,

much attention was given the areas of motivation, learn­

ing, emotional state, etc., as well as the effects of

temperature and himiidlty, altitude, noise, vibration,

illumination, radiation, and other environmental factors

on the human organism. All of this research has essen­

tially tha same goal, to maximize man's efficiency in the

performance of a task^

Many researchers felt that this goal could be

achieved by simply minimizing the amount of physical

effort or the energy expenditure of the human operator.

However, before physical effort can be minimized, one

must be able to measure It, Just what is "little physical

effort," and what is "considerable physical effort"? It

seems indeed understandable that so many attempts have

been made to find some measureable dimensions for human

energy expenditure that may permit a better rating and

understanding of this component of industrial work^ (19)

Almost every technique of "technical measurement"

may be classified in one of four categories;

1. Straight output measure

2. Fatigue measurements

3. The mechanical-physical system

4. The physiological system

Straight output measure is the system known as piece work. While it never actually claimed to be a measure of effort or energy expenditure, high piece work performance too often became synonymous with high effort. However, the mere

count of the number of pieces produced by the worker during the day does not tell us very much about energy expenditure. It has no application at all to the results of effort which due to their nature cannot be expressed by time con­sumption. (19)

Fatigue measurements and the Introduction of

fatigue allowances were the first step to understanding

the complex human contributions to some counted output.

Many industrial engineers thought this was the answer to

their problem^ Such enthusiasm was somewhat hasty since

mainly jobs with a fairly high degree of physical exertion

had been examined and these jobs are rare in modern In­

dustry. These early fatigue measurements were made by

counting the number of pieces produced during consecutive

hours of the working day and drawing fatigue cur'ves. (19)

It soon became apparent that under conditions re­

quiring less than near-to-maximum physical effort, the form

of such curves was determined more by incentive or motiva­

tion than by fatigue. This approach actually tended to

hide the real amount of individual energy expenditure.

Due to its many subjective aspects, fatigue cannot be con­

sidered a good measure of expended energy. (19)

The mechanical-physical system makes use of the

basic laws of mechanics. Numerous attempts have been made

to fit the principles to human v/ork However, mechanical

work formulas must be modified before they can be applied

to human tasks. The problem is one of finding some

7

additional factor that makes sense to the "human motor".

As yet, we do not know what this factor might be. (19)

We now come to the category of technical measure-

ment known as the physiological system. Physiological

measurements represent a more direct attack on the

phenomena of energy expenditure in h-uman work.

The CO2 method represented one of the earliest

approaches to physiological measurements. An attempt was

made to obtain reliable knowledge about human energy

expenditures by measuring the amount of oxygen converted

into carbon dioxide. Extremely valuable information

concerning the energy efficiency of certain prototype

jobs was provided by this method, (19)

As more sophisticated measuring devices became

available, additional methods of determining the "physio­

logical cost" to the human being in the performance of a

task came into use. Some of the more common and widely

used of these physiological indices of performance are?

1. Metabolic rate 2. Heart rate 3. Blood pressure 4. Blood flow rate 5. Pulmonary ventilation 6. Oxygen consumption 7. Carbon dioxide production 8. Rate of perspiration 9s Skin temperature 10. Electromyography

All of these techniques have certa.ln dis"f""'nrt dis­

advantages in that most require the attacl iment of apparatus

8

to the subject, some require precise laboratory analysis,

and others are just inherently difficult to measure.

Some techniques reflect all of these disadvantages.

The acceptance of heart rate as a measure of

physiological cost was based on the assumption that a

faster heart rate is associated with more effort or energy

exerted. Recently, however, it has been found that there

is not a linear relationship between the energy exerted

and heart rate. In one study subjects pedaled a bicycle

ergometer at 30 rpm for several hours, and it was found

that the heart beat rate rises for approximately 30 min­

utes after the start of work even though a constant amount

of energy is being exerted. After 30 minutes, the heart

rate was again relatively constant. (21)

The oxygen consumption method, which has been the

most widely accepted approach, requires the attachment of

apparatus to the subject which in itself may well distort

the interaction between the subject and the work situation.

(18)

In the carbon dioxide method the worker is required

to wear a mask which has a physical as well as a psychologi­

cal effect on his performance. "Another problem is that the

oxygen 'debt' Incurred has a time lag which makes it diffi­

cult to assign a quantitative value to energy used (ex­

pended) on a specific part of the task (21)•"

Many of these measurement problems can be overcome

by the use of a fairly recent device for measuring physio­

logical cost--the force platform. Since this device has

been selected for use in this particular experiment, let

us examine it in greater detail.

The Force Platform

This device was developed by Ls Lauru in 1953^

improved by Greene and Morris (1959), and further modi­

fied by Barany (1961) with the aid of a grant from the

National Science Foundations It is capable of detecting

minute forces in three independent perpendicular planes.

The subject stands on the platform and there is no other

contact between the subject and apparatus.

Movements of the platform are detected by a

pressure-sensing device, such as a strain gauge or

plezioelectrlc crystal, for each of the frontal, vertical,

and lateral planes. Independence of the three axes is

assured through the use of an equilateral triangular

support of the vertical forces and by single point trans­

mittal for the lateral and frontal forces (21). Greene

and Morris (1959) demonstrated the fact that resolution

of forces into the three planes of motion is possible

with this type construction (12)s Hudson (1962) examined

the dynamic characteristics of the frequency responses

and found them to be inconsequential (l8)^ He makes the

10

following statements concerning his investigation:

The general conclusion as of now is that the evaluation did not reveal any characteristic which constitutes a serious obstacle to use of (this) ap­paratus as a primary experimental instrument. The platform.•.was subjected to an extensive series of tests and the force response curve for all axes is linear within very narrow limits. There is no evidence of interaction between axes. The sensi­tivity exceeds all experimental requirements. In fact the platform appears sensitive enough to detect tha human heart beat of the subject. (l8)

The output of the force platform can be recorded

by either an ink-supplied, pen type stylus on regular re­

cording paper or by a resistance type stylus on special

heat sensitive paper. The force trace originates from an

established zero mark and deviates up or down. The dis­

tance deviated is directly proportional to the force ex­

erted in the specific plane. According to Konz and Day

(21), two analysis techniques can be used:

1, The total area recorded (i.e. energy) can be calculated either by using a planimeter or by feeding the signal into an analog integrating circuit.

2. The maximum height of the trace can be used as an index of the forces exerted by the operator when performing the task. Barany (1963) demonstrated that using the maximum height is sufficient for some tasks. In his experiment he found a high correlation between the area under the force trace and the maximum height of the force trace. Therefore, if the energy exerted is required, the area analysis should be used; if only the maximum force exerted is required, measuring the maximum height of the line is sufficients

11

A third method of analysis was described by Whit­

ney (28) in 1958 in an experiment using the force plat­

form to determine the strength of the lifting action in

man^ He wanted to determine the average deflection of

the central two-thirds of each record from Its respective

baseline. The central portion was considered more repre­

sentative of the subject's performance under the specified

conditions. He measured the area under the central 2 cm

of each record with a planimeter and then divided this

area (in cm ) by 2 cm. This average deflection in cm was

then converted into its equivalent force by means of the

calibration record of the platform obtained at the con­

clusion of each series of trials. Figure 1 illustrates

the use of this methods The shaded area A represents the

central portion previously mentioned• The value of c

is obtained from the calibration record of the platform •

The use of the force platform as a measure of

physiological cost was justified when an investigation by

Greene (1957) of the relationship between two widely ac­

cepted measures of physiological cost and the force trace

records of the platform indicated that the force traces

compare favorably with the two other measures when the

trace is properly interpreted• In fact, he suggested ttiat

the force platform method is superior to both the oxygen

consumption method and pulse rate method as a measure of

the physiological cost of certain types of dynamic work, (13)

A = 10 cm'

>2 _ Average deflection = 10 cm^ 5 cm

c = 4 lb per cm 2 cm

Average force = 5c = 5(4) — 20 lb

12

Flg^ 1•--Whitney's method

13

Purpose and Scope

The purpose of this experiment is to determine the

affect of certain operational and design parameters on

tha bodily reaction forces developed by an operator of a

handwheel used in a strenuous dynamic task and to deter­

mine the optimum values of these parameters, thereby pro­

viding information of great use in the design and/or oper­

ation of handwheel controls• The specific parameters to

be evaluated are:

1• The diameter of the handwheel 2. The resistant torque that must be

overcome 3^ The mode of operation (i^e^, place­

ment of the hands on the wheel)

4, The height of the operation

The results of this study should also enable conclusions

to be drawn concerning optimiim combination of these para­

meters for similar tasks involving the application of high

torque.

"By far the greatest amount of work on handwheels

has been to examine the influence of various factors on

their effectiveness as tracking controls." (4)

The measure selected to evaluate handwheels used in

tracking tasks has typically been one such as accuracy or

maximum turning rate, etc. Little attention has been given

to the evaluation of handwheels used for tasks other than

tracking, such as the application of torque in order to

open a valve or move a weight through a distance, and the

14

measure of physiological cost to the operator has been

largely neglected• Also, low values of resistant or

braking torque have been used in most of the previous

experiments•

In this experiment, handwheels will be evaluated

for a task other than tracking in which, for the most

part, high values of resistant torque must be overcome•

Tha output of the force platform will be used as a measure

of the physiological cost to the operator.

Before proceeding further, let us examine some of

the previous research in the areas of handwheel design

and operation•

Review of Previous Research

One of the most extensive investigations of hand-

wheels was made by Davis (6) in a two-part study performed

in 19^9 and 1951. In the first part of the study he intro­

duced the following variables: size of the handwheel,

frictional torque applied to the handwheel shaft, and

location of the handwheel in relation to the operator both

as to height and angle of the axis of the handwheel shaft.

The task the subjects were required to perform was the

setting of an indicator to a specific position by turning

the handwheel one complete revolution.

The handwheel diameters evaluated were 3 in^, 6 in.,

8 in^, 10 in., and l6 in^ The values of torque used were

15

0, 20, 40, and 90 inch pounds. The desired torques were

achieved by applying friction loads to the handwheel shaft

by adjusting C-clamps around two oak vise jaws surrounding

the shafts Performance was evaluated at heights of 24,

36, 39, ^0, 42, 48, and 58 inches above the floor, and

with the axis of rotation of the handwheel at 0 degrees

(perpendicular to the sub ject), -f 45 degrees, -45 degrees,

and 90 degrees (parallel to the subject).

Davis found that:

For each control location there is a breaking point or a divergence point at which the relation­ship between size of device used and performance changes. This divergence point is usually at the 40 Inch-pound torque. Above this point, the larger the control device, the better the performance achieved; below this point, the smaller the size of control device, the better tha results• Above the divergence point, the larger the frictional torque, the larger the difference grows in favor of the larger devices. (6)

Table 1 summarizes the more important results of

Davis' investigation with regard to optimum sizes of hand-

wheels. It must be remembered that these findings are based

on one revolution of the handwheel and that the operator

turned them with one hand by means of a handle on the wheel

rather than by using the rim.

In part two of Davis' study (1951), be Investigated

the same variables of height, angle, torque, size, and type

of device as in the prior experiment, but this time he

required that the operators make the setting in one-half

revolution. He found that:

16

TABLE 1

OPTIMUM DIAMETER OP HANDWHEELS, IN INCHES, FOR VARIOUS TORQUES AND POSITIONS (FROM DAVIS)

Height from

floor, inches

36

36

40

42

Position degrees*

0 (front)

0 (side)

- 45

-f 45

Torque, inch-pound • •

0

3,6,8,

3,6

3,6

3.6

20

10,16

10

10,16

6,10

40

10,16

10

6,10,16

10

'. !• . ' • t

s

90

16

10

10,16

10,16

^Degrees from horizontal of shaft of wheel

17

For small control devices such as 1^ and 3 inch radius wheels...best performance is obtained at minimum torque for all locations. Any increase in torque above zero inch-pound results in a mark­ed reduction in performance. Performance of the intermediate size wheel, 5 in. radius, is only slightly influenced by torque. Optimal performance for all locations comas at 20 to 40 inch-pound-tor­que. However, performance at these torques is not greatly different from that at the highest or low­est torques. Regardless of the,,.size of the con­trol device or the torque at which it is operated, better performance is obtained at the 40 inch height, minus 45 degrees location than at 36 inch horizontal axis which can be considered as typical. (6)

Halson (19^9) investigated the performance of oper­

ators using handwheels in a tracking task and concluded

that: 1, A large-sized wheel Is consistently super­

ior to a small-sized one up to 100 rpm when the advantage passes to the latter.

2. Accuracy of tracking increases with speed of turning.

3^ Past turning speeds are superior for pro­longed as well as short periods of track­ing.

4. The advantage of heavy over light hand-wheels at all tracking speeds is clearly evident. (15)

In this experiment the operators were seated and

the handwheels were either 28,5 or 30 Inches from the

floor and l4 inches to the right of the target. Runs

lasted either 3 or 4 minutes.

The Poxboro Company conducted a number of experi­

ments to determine the effects of orientation, inertia,

friction, and diameter of the wheel on tracking accuracy

(l4). The display device consisted of two black pointers

on a white scale, surrounded by a gray matte surface. The

18

bottom pointer served as a stationary zero point. The top

pointer was controlled by a metal wheel, 6 Inches in dia­

meter, with a 3 inch wooden handle at a right angle to the

plane of the wheel. Tha wheel center was 28i Inches from

tha floor and 24 inches from the subject.

The following conclusions were drawn from the results

of these experiments:

1. There is no permanent advantage in using either horizontal or vertical position of handwheels for tracking performance.

2. Little effect on accuracy was traceable to the difference in position of the hand-wheel (vertical or oblique),

3. Inertia reduced tracking error materially and had a marked smoothing effect on per­formance .

4. The difference between the 4J and 9 inch diameter sizes did not make for large dif­ferences in accuracy or smootv ness of tracking, except that with friction, the larger wheel was better.

5. Friction is definitely undesirable. As the affective frictional torque at the handwheel increases, performance gets rapidly worse, especially at low speeds. (14)

In an experiment performed by Katchmar (1957), sub­

jects cranked handwheels of different radii and loads for

varying periods of time. Seventy-five subjects and three

wheel radii (4, 5, and 7 in.) were used. The loads ranged

from zero to 90 inch-pounds and the cranking time varied

from one to ten minutes. Results Indicated that the loca­

tion and orientation of handwheels make little difference

in cranking speed and accuracy as long as they can be oper­

ated comfortably. (5)

19

Morgan (25) says:

1. The diameter of the handwheel rim should not exceed 3A "^ 2 inches.

2. For most effective use, handwheel displace­ment should not exceed plus or minus 6o degrees from the normal (null) position be­cause larger arcs require the hands to shift position on the control.

Corlett (l4) conducted a preliminary experiment in

1961 to examine tha effects of certain factors that he con­

sidered relevant to the problem, and to determine which

should be examined in greater detail. The factors consid­

ered relevant were:

1. Handwheel diameter

2. Dial diameter

3. Line thickness

4. Spindle friction

5. Working height

Results of the survey experiment Indicated tnat the most promising fields for future experimentation would be tha variation of (a) dial diameter and line thickness, and (b) handwheel diameter and friction on the spindle. The value at which the other listed factors should be held constant were shown to have little effect on the results, (l4)

Lehman (23) reports on a study of a considerably dif­

ferent nature. One of the purposes of this study was to

determine tha most suitable location of the steering wheel

of a tractor from an investigation of the energy cost of

steering and the force and speed achieved. The techniques

employed were measurement of oxygen consumption and heart

rata of the operator and of forces required in turning the

20

steering wheel. Lehman summarizes the findings of the

study as follows:

The largest force can be exerted on an almost horizontal steering wheel. On the other hand, the steering wheel may be turned with the greatest velocity if it is nearly vertical. With this vertical position of the wheel, however, the energy consimiption is very high, and is smallest with the steering wheel inclined so that its axis (i.e., the steering column) makes an angle of 50 - 6o degrees with the horizontal. In this range only 70 per cent of the maximum force is achieved but this position of the steering column must be considered the most favorable one, (23)

These findings must be viewed in the light of the

fact that only one (unspecified) wheel size was used and

the effects of varying the heights and frictional torque

were not investigated.

By now it is apparent that additional research con­

cerning handwheel controls is warranted and desirable.

Research involving non-tracking applications is especially

needed. It is this apparent need that stimulated research

culminating in the experiment treated in the succeeding

chapters.

CHAPTER II

EXPERIMENTAL PROCEDURE

Subjects

The subjects in the experiment were volunteers from

the male population of the university. Five volunteers

whose schedules suited the experiment were selected. The

subjects selected were of similar build and thus consti­

tuted a relatively homogeneous group with respect to

anthropometric measurements. All subjects were checked

to Insure that they had no injuries to hands or arms.

Their personal data relevant to this research appears in

Table 2. The average age as shown by Table 2 was 26.2

years. The standard deviation was less than one year

(0.955) so age can be disregarded as a factor in this

experiment. All subjects were right-handed.

Apparatus

The following equipment was used in the experiment:

1. A Beckman Dynograph Recorder

2. Force platform

3. Torque device

4. Metronome

The Beckman Dynograph Recorder (see Figure 2) is

a highly sensitive oscillograph capable of simultaneously

recording signals in different modes from many sources,

21

22

CVI

<

<

< p tH O

CO

4J -^^ ^ P . O bOC P -H .H M 0'-—• W P

PH 0 4J^-v T S P • M bO C ps -H -H O 0 * - ^ P P CO

o 4.> 0

M ^ - s ^ M . O O PI P P5 tH M P I ^ - -W W

P. ^ 0 o ^ - ^ TS P • M M Pi ps W - H o - P o CO 4^

4:> P ^ - ^ bOP •H M 0 :s

4 J y - s ,

P • bOPi •H •H 0 ^ ^ P

0 bO <

4^ o 0 •

•r-3 O p : ^ : ps CO

O •

OJ ^

o • i n i n

l A ,

.=j-r-{

o »

cn M

CO !>-M

i n .

t ^ VD

I -CM

M

O •

LA ^

O .

!S-i n

i n ,

i n M

o • OJ M

o 1--H

o , M t -

i n CVJ

CVJ

i n .

cn ^

o 0

VD i n

o .

i n M

i n ,

CM M

O 00 M

o •

a VD

i n CM

(Y^

o ,

i n .:^

i n .

00 i n

o VO i H

i n •

cn M

00 i ^ H

o o

M l ^

o CY

.=t

o .

Cvl ^

o ,

^ i n

i n ,

^ M

'" '

o • CM M

CM t>-M

O .

CO VD

^ CM

i n

23

Pig . 2.--Beckman Dynograph Racordar

24

It was used to amplify and record the three outputs of tha

force platform.

The force platform shown in Figure 3 was used to

detect the bodily reaction forces produced in performing

the experimental task. This force platform was construct­

ed at Texas Technological College using a design based

on one developed principally by Barany and Whetsel of

Purdue University in conjunction with a National Science

Foundation grant. Their platform is a redesigned version

of the original model constructed by Greene (12), The

new design maintains the geometric properties of the

equilateral triangular placement of cantilever beams and

tha utilization of Linear Variable Differential Trans­

formers (LVDT's) in the sensing units but possesses the

added features of being compact, portable, and relatively

inexpensive to reproduce. The device weighs less than

100 pounds, with overall dimensions of 25" x 22" x 5".

Mechanically, the force platform consists of a 3/4

inch thick hexagonal aluminum top-plate and truss section

suspended vertically by six horizontal cantilever beams

and restricted horizontally by six vertical cantilever

beams. In turn, the twelve beams plus three sensing units

are supported by an aluminum base plate. The schematic

diagram In Figure 4 illustrates the arrangement of the

beams and sensing units.

Forces VI, V2, and V3 in Figure 4 represent the

25

Pig. 3.—Force platform

26

^ i g . 4 . _ orc

FRONTAL

' Platform schemati.

27

weight of the suspended portion of the platform plus any

other downward vertical force that might be exerted on tha

instrument. These three forces are exerted at three points

which form the vertices of an equilateral triangle. Forces

V4, V5, and V6 represent the upward vertical forces that

might be exerted on the platform during transportation or

large displacamants. The three upward forces form the

vertices of a second equilateral triangle superimposed on

the triangle formed by the downward forces. The common

intersection of the perpendicular bisectors of the sides

of the triangle formed by the upward forces is on the same

vertical axis as the intersections of the perpendicular

bisectors of the sides of the triangle formed by the down­

ward forces. This arrangement not only restricts the plat­

form in both vertical directions, but, a constant force

exerted in either direction anywhere on the platform will

result in the same amount of deflection at the center point

of the hexagonal top-plate•

Forces Fl, P2, P3, and F4 represent the frontal

forces and forces LI and L2 represent the lateral forces

exerted on the platform. Horizontal movement of the sus­

pended portion of the platform is restricted by the two

sets of vertical beams shown in Figure 4,

The restrictive forces in the vertical and horizontal

directions are exerted at point contacts consisting of steal

balls pushing against flat tool-steel surfaces attached to

28

the cantilever beams. Therefore, only forces perpendicular

to the restricting beams will cause beam deflection.

Deflection in the three orthogonal directions is

detected by shielded Linear Variable Differential Trans­

formers (LVDT's). These devices consist of three adjacent

colls wound on the same insulated spool. A magnetic core

is initially located in the center of the middle coll.

Linear deflection of the core along the axis of the colls

causes increased coupling of the center coil with one of

tha end coils while linear deflection in the opposite

direction causes Increased coupling of the other end coil.

This voltage differential in the end coils is used as an

input signal to the Dynograph Recorder.

The three sensing units SI, S2, and S3 are located

as shown in Figure 4, The vertical deflection sensing

unit SI is placed at the common center of the two equi­

lateral triangles. An adjustable set screw through the

center brace of the truss section is used as the sensing

face. The frontal sensing unit S2 is placed along the

frontal axis passing through the center of the platform,

A vertical plate perpendicular to the frontal axis is used

as the sensing face. Only deflections parallel to the

frontal axis will displace the core. The lateral sensing

unit S3 is placed along the transversal axis which passes

through the center of the platform• A sensing face simi­

lar to the frontal one is used so that only deflections

29

parallel to the lateral axis will displace tha core in the

lateral sensing unit.

A locally manufactured device was used for producing

the desired values of resistant torque (20)• Figure 5 shows

a sketch of the device• It is essentially a double block

brake. The blocks are made of wood with leather bearing

surfaces; the drum is made of metal and is fixed to a

central shaft. When a weight is suspended from the cable

the two wooden blocks are forced against the drum causing

a constant braking torque to be applied to the central

shaft. The device was clamped to a surface which could be

raised or lowered to the desired height. A picture of the

device is shown in Figure 6.

A metronome was used to pace the experimental task.

Pacing the task Insured that, for any given trial, all

subjects turned the handwheel at relatively the same angu­

lar velocity.

Task

The task that the subjects were required to perform

was a simple, dynamic manual task such as that performed in

opening a valve or loosening a vise. In this type of task

high initial resistant torque must be overcome during a par­

tial revolution of the handwheel. The subject, while stand­

ing on the force platform, grasped the handwheel by the rim

with both hands and turned it counterclockwise through an

30

Pulley- Upper -1 Arm

Drum Shaft

Lower Arm

Upper Block

Lower Block

Pig* 5.—Torque device schematic

31

P ig . 6.--Torque device

32

arc of 90 degrees. This value was arbitrarily chosen but

is appropriate for the task and does not exceed the maxi­

mum displacement of 120 degrees recommended by McCormick

( 2 2 ) .

Performance of the Experiment

The entire experiment was performed in an air con­

ditioned research room where interruptions could be held

to a minimum. The experimenter and recorder were placed

at 90 degrees to the subject to minimize distractions.

Before beginning the experiment, the recorder was

turned on and the torque device was calibrated by means

of a torque wrench which was turned in time to the metro­

nome. The metronome was set at the same setting as that

at which the experimental task was paced during the actual

experiment. This insured a relatively constant resistant

torque for this particular speed of rotation, which was

used throughout the experiment.

After the recorder had warmed up sufficiently, the

force platform was then calibrated. Calibration of the

platform for vertical forces was accomplished by placing

standard weights on the platform and noting changes in pen

deflection as a function of Increased load, A calibration

table utilizing pulleys was necessary in checking the meas­

urement of horizontal forces. This device is shown in

Figure 7. Calibration records for the force platform

33

Fig. 7.--Calibration table

34

appear in Appendix A^

After calibrating the platform, the subject was

brought into the experimental room and given a thorough

briefing on the nature and conduct of the experiment.

After this had been done, the subject was directed to

stand at the center of the platform with his heels six

inches apart and feet aligned with lines on the platform.

This standard standing position insured that the verti­

cal force due to the subject's weight passed through the

center of rotation of the platform. The subject stood in

a relaxed position on the platform with his arms hanging

naturally at his sides.

The specified diameter handwheel for the trial was

mounted on the shaft of the torque device which was affixed

to an adjustable stand resting on the floor. The stand was

positioned so that a fixed distance of 15-3/^ Inches as

measured from the acromion process at the shoulder of the

subject to a vertical plane passing through the midpoint

of the longitudinal axis of the handwheel hub was maintain­

ed. The layout of the equipment and the experimental work­

place are shown in Figures 8 and 9.

The torque and height were then adjusted to those

values required for the particular trials The recorder was

turned on for a few seconds to record the normal weight

reading for the subject and the recording pens were adjust­

ed to a null position representing a zero activity levels

35

Flg^ 8.--Torque device and force platform

36

Pig. 9.—Experiment In progress

37

With the recorder off, the subject was told to

grasp the handwheel by the rim with both hands, placing

his hands in the positions for that trials The metronome

was turned on and the subject was given practice in rotat­

ing the wheel through 90 degrees, regrasping in the same

position, rotating again through 90 degrees, etc, in time

with the metronome. When the experimenter was satisfied

that the subject was sufficiently trained and the hand

placements were correct for the trial, the recorder was

turned on and the pens re-zeroed, if necessary. The sub­

ject was then told to grasp the wheel, turn it counter­

clockwise through the prescribed arc, regrasp, and turn

again In time with the metronome until the wheel had been

turned five times. The metronome was set at 40 beats par

minute throughout the experim.ent. Upon completion of the

required number of turns, the subject returned to the

"relaxed position" where his null reading was again re­

corded. The recorder was then turned off and preparations

made for the next trial.

The null or zero line established bench marks both

before and after the task which reduced the problem of any

electronic drift of the equipment as well as other mis­

cellaneous errors. The subject standing with his arms at

his sides was used as a baseline for this experiment as the

output may then be Interpreted as the total force required

for the complete task (above that required to stand in

38

place).

Each subject performed ninety-six trials repre­

senting the ninety-six possible combinations of height,

torque, etc. After every five trials, the subject was

given a sufficient amount of time to rest in order to

minimize the chance of fatigue affecting the results.

The time-order of performance for each trial was deter­

mined by a table of random numbers.

At the end of each series of trials the subject

was asked which hand position or positions he preferred

at each of the heights. Also at this time the calibra­

tion of both the force platform and torque device was

checked again. If the readings at the beginning and end

had been significantly different, the series would have

been rerun.

CHAPTER III

DESIGN OF THE EXPERIMENT

Selection of Variables

The variables considered relevant in an experiment

of this type fall Into two categories:

1^ Variables that are a function of the differences between subjects

2. Variables that are a function of the control design and/or workplace layout

The first category includes:

a. Subject height b. Subject weight c. Shoulder breadth and height d. Arm length a. Lateral location of the control with

respect to the subject f. Fore-and-aft location of tha control g. Maximum strength of the operator h. Susceptibility to fatigue 1, Mode of operation (hand positions) j• Spaed of rotation

The second category Includes such factors as:

a. Diameter of the handwheel control b. The resistant torque that must ba over­

come 0, Vertical location d^ Control orientation a• Direction of movement f, Amount of movement

The affects of the first four variables in the

first category were controlled so that their influence was

negligible by selecting a relatively homogeneous sample of

subjects and using subjects as blocks in a randomized

block design.

39

40

Lateral location was fixed. The control was lo­

cated directly in front of the operator's midline. This

is the recommanded location for positioning movements in­

volving both hands as given by Briggs (5).

For rotary movements there is little difference in

the forces exertable at most fore-and-aft locations,

though the farthest points should be avoided (5). This

being the case, a fixed distance of 15-3/^ inches as

measured from the acromion process at the shoulder to a

vertical plane passing through the center of the longi­

tudinal axis of the handwheel hub was used since this

would accommodate very close to 100 per cent of the popu­

lation (24,8 inches will accommodate all but one per cent)^

The maximum strength of the operator was not a

factor in this experiment. Each subject was tested be­

fore experimentation using the most disadvantageous com­

bination of resistant torque and handwheel size to insure

that the force required to turn the handwheel did not ex­

ceed the maximum strength of any subject. Resistant torque

did not exceed 120 inch-pounds and handwheel diameter was

no smaller than 7 inches. For this combination the com­

bined force that must be exerted by both hands is less

than 35 pounds.

Subjects were given frequent rest periods of ade­

quate length in order to eliminate fatigue as a source of

variation. Although some subjects are more susceptible

41

to fatigue than others, by closely observing the force

traces the experimenter can recognize its presence since

the recording st:jrlus will begin to make sudden rapid de­

viations from the zero activity line while the subject is

in a relaxed position on tha platform. The subject will

be unable to maintain the stylus on the zero line due to

loss of muscular control, especially over those muscles

controlling posture and balance. This phenomenom has

been observed many times by this experimenter In prior

work with the force platform. Had it been observed in

this experiment, the subject would have been given an

unscheduled rest period.

Speed of rotation of the handwheel was maintained

relatively constant for all subjects for any given trial.

This was accomplished by having the subjects perform the

task in time with a metronome. Subjects were given ade­

quate training in timing their movements before performing

each trial.

The last three factors in the second category ware

also controlled. One fixed control orientation was used

in this experiment. The handwheel was oriented so that

the axis of rotation was perpendicular to the frontal

longitudinal plane of the subject's body^ This was the

only orientation feasible due to the limitations of the

torque device.

Direction of movement was fixed. The handwheel was

42

rotated in a counterclockwise direction which is the most

realistic for the task chosen.

The handwheel was rotated through an arc of 90 de­

grees. This value does not exceed the maximum recommended

value of 120 degrees and, again, is realistic enough for

the particular t^sk that was performed.

The four remaining independent variables are the

ones that were chosen for consideration in this experiment.

They are:

1, Handwheel diameter

2, Vertical location

3, Resistant torque

4, Mode of operation (hand position)

Handwheel diameters of 7, 10, and l4 inches were

selected (see Figure 10). These are sizes frequently

encountered in industry and the size Increments are such

that if there is a breaking point or divergence point on

the torque scale at which the relationship between size

of device and performance changes, the probability of de­

tecting this point is relatively high.

Two relative heights were used. One was six inches

above the elbow height of the subject and the other six

inches below• The lower height would correspond to a con­

trol height of 36 Inches for an average subject. Hand-

wheels used in industrial applications are frequently found

at this height. The second height was different enough from

43

Pig^ 10.--Handwheels used

44

the first that it enabled the effect of vertical location

to be more readily determined. Having these particular

heights enabled a determination to be made as to whether

location above or below the elbow is more advantageous.

Resistant torques of 20, 40, 80, and 120 inch-

pounds were applied to the handwheel shaft. Here again,

the increments were large enough that they would have re­

flected any significant effect produced by varying the

torque. One of the reasons why high torque values were

selected is that they are representative of the task being

evaluated in this experiment. Also, these high torque

values required the application of forces of large enough

magnitude such that they did not demand that the sensitiv­

ity of the recorder be as high nor as critical as in tasks

where smaller forces are applied. This, in turn, promoted

greater reliability since a small error in measurement

would not have appreciably affected the resultant force;

whereas, in the case of small applied forces, a small er­

ror would have precipitated a completely erroneous inter­

pretation of the force trace•

The subject grasped the rim of the handwheel in four

different ways^ Consideration of hand placements as a legi­

timate variable was justified by a trial run which indicated

that this factor does Indeed affect the force platform out­

put. Tha various hand placements can best be explained in

conjunction with Figure 11, They are:

45

0 Mode 1:0= 60°

R ® Mode II: e = 60°

R

Mode III: 0 = 120° Mode IV: 9 = 180°

Fig^ 11•--Hand positions

46

(a) Mode I—Left hand at ten o»clock position and right hand at twelve o'clock position.

(b) Mode II--Left hand at twelve o'clock position and'right hand at two o'clock position,

(c) Mode III--Left hand at ten o'clock position and right hand at two o'clock position.

(d) Mode IV--Left hand at nine o'clock position and right hand at three o'clock position.

Each of the ninety-six possible combinations of

the four variables selected was run in a random order for

each subject. In every case the dependent variable was

the force platform output.

Discussion of Variables Selected ^ ' I I I ' I I Ill » * B l l l ^ l H f • • I I P . • • • ^ 1 W m I I I . ' W ^ . 1 1

It was anticipated that the effect of increasing

the handwheel diameter would be predictable up to a pointy

Increasing the diameter increases the moment arm of the

force that must be applied to overcome the resistant tor­

que. For the same value of resistant torque. Increasing

the diameter decreases the amount of applied force requir­

ed. In general, the larger the frictional torque, the

larger the handwheel size would have to be in order to

lessen the physiological cost to the operator.

Table 3 clearly illustrates these facts. Of course,

there is a limit to how large the handwheel can be before

it exceeds the capabilities of the operator and/or becomes

TABLE 3

47

APPLIED EQUILIBRIUM FORCE IN POUNDS FOR VARIOUS TORQUES

AND HANDWHEEL SIZES .-- , .' 1 -

Torque in

inch-pounds

20

40

80

120

7

5.71

11,43

22.86

34.29

Handwha ( Diamate (inches

10

4.00

8,00

16.00

24.00

el r )

14

2.86

5.71

11.43

17.14

(Forces Just greater than those listed will cause movement of the handwheel)

48

more costly to the operator. If the handwheel diameter

becomes too large, a greater number of muscle groups and

of different type than at the smaller diameters are brought

into play. The sizable moment arms created by very large

handwheels require the use of the large muscles of the

back and trunk. This tends to rotate the entire body and

to greatly displace its normal center of gravity, creat­

ing a condition of Imbalance. Great reactive forces in­

consistent with the task evaluated here would then have to

be applied by certain muscles of the body in order to

maintain balance. To avoid this condition and the possi­

bility of introducing additional variables, handwheel

sizes large than l4 inches were not considered. This would

Insure that major muscle activity would ba confined to rela­

tively the same muscle groups, primarily those of the arms

and shoulders.

Increasing the diameter and, in effect, the moment

arm of the applied force, should Increase the bodily reac­

tion forces to some extent for those handwheels that are

not inordinately large. In particular, an Increase in the

transverse forces should be apparent. The question of

primary Interest here is, is there a divergence point above

which the advantage lies with the larger devices and below

which the smaller devices are more advantageous? Davis (6)

found such a divergence point which he says is usually at

the 40 inch-pound torque. However, the task which he

49

evaluated was considerably different from the task eval­

uated here. There is no guarantee that such a point exists

or, if it exists, can be readily determined for a task in

which accuracy is not a factor and is not the measure of

performance being evaluated.

Increasing the resistant torque will, for the same

diameter wheel, require that the applied force be Increased,

In general, the lowest value of frictional torque will pro­

vide the least physiological cost to the operator. What

is Important here is not the resistant torque main effect,

per se, but the interaction of resistant torque with the

other main effects. Of particular interest is the inter­

action with handwheel size. This Interaction should pin­

point the divergence point previously discussed if such a

point exists. Although the resistant torque was not of

primary interest, it was expected that it would be highly

significant.

A number of studies have been performed in an at­

tempt to determine optimimi heights for certain tasks.

However, the majority of these studies were concerned with

horizontal work surfaces. Ellis (8) performed an experi­

ment involving a manipulative task by persons working in a

standing posture and found a level around 3 inches below

the elbow to be optimum. This was, on the average, about

42 inches above the floor. From this study and from the

experience reported by Barnes (3), there is a substantial

50

basis for arranging work surface height for standing work­

ers at a level somewhat below elbow-level height (about 2

to 4 inches), at least for light assembly work or similar

manipulative tasks. Where movements involving consider­

able exertion are required, somewhat lower levels probably

would be in order (24). These findings were not expected

to be necessarily applicable in this experiment involving

a vertical work surface. It should also be noted that the

effect of height is closely related to the arm length and

shoulder height of the subject. For any particular sub­

ject, fixing the fore-and-aft distance and the control

height will completely determine the elbow position (height)

and angle of flexion. The optimum height was expected to

be the one for which the angle of flexion of the elbow was

closer to 90 degrees. The value of 90 degrees was sub­

stantiated in an experiment by Provins and Salter as being

the optimum angle of elbow flexion for a task requiring a

relatively high degree of exertion (24).

The effect of the mode of operation (hand position)

is not easily predicted. Static blomechanical analysis

becomes very difficult in short order and is unrealistic

at best. Dynamic analysis, which is more realistic, pre­

sents the problems of measuring such quantities as angular

displacement, time, length, weight, and weight distribu­

tion. A prohibitively large number of kinematic quantities

and physical constants must be determined by measurement.

51

Also, very sophisticated techniques of measurement are

required. However, some hypotheses concerning mode of

operation were made based on a little logic, experience,

and intuition. This experimenter felt that the more

balanced hand placements would be less costly from a

physiological standpoint. These positions enable forces

to be applied more smoothly and with greater control.

Positions where both hands are on the same side of the

null position would cause an unbalanced force to be ap­

plied which would tend to cause a shift in body weight.

A corresponding reactive force would be applied by cer­

tain muscles of the body in an effort to maintain balance,

As the angle between the hands in these unbalanced posi­

tions becomes smaller, the bodily reaction forces that

must be applied become larger•

The preceding comments may be summarized in the

following Initial hypotheses that were made prior to ex­

perimentation:

1. Larger handwheel sizes are more advan­tageous at higher torque values requir­ing large applied force.

2. Lower torque values are less costly,

3. Vertical locations for which the elbow angle is close to 90 degrees are better for handwheel controls•

4. More balanced hand positions are less costly.

The first two hypotheses were dictated by common

52

sense and are readily apparent. The latter two may not

be so apparent and were based on preliminary analysis and

experience.

Design of the Experiment

A four by four by three by two factorial experi­

ment In a randomized block design was used. Each subject

was treated as a block with a complete factorial experi­

ment randomized within each block.

The mathematical model of the experiment is:

Where:

%1klm ~ Resultant bodily reaction force

u = A common effect in all observations

SjL = Subjects (blocks)

D^ = Handwheel diameters

Hj^ = Heights

T-j_ = Resistant torque

Pj^ = Hand positions

eijklm = Random error in the experiment

1 = 1, 2, 3, 4, 5 1 = 1 , 2, 3, 4

j = 1, 2, 3 m - 1, 2, 3, 4

k = 1, 2

53

The expected mean squares (EMS) and degrees of

freedom (df) are given in Table 4, These EMS values in­

dicate that all main effects and interactions can be

tested against the 380 df error term. All of the inde­

pendent variables considered In the experiment are sum­

marized in Table 5,

TABLE 4

EMS FOR FOUR-FACTOR EXPERIMENT IN A RANDOMIZED BLOCK DESIGN

54

1

Source

h ^i

" k

T l

Pm

^'^Jk

""hi

^P jm

HTkl

I^^km

TPim

DHTjkl

jkm

^•^^klm

^ ^ ^ ^ j k l m

® i j k l m

T o t a l

d f

4

2

1

3

3

2

6

6

3

3

9

6

6

18

9

18

380

479

5 R 1

1

5

5

5

5

5

5

5

5

5

5

5

5

5

5

5

1

3 P J

3

0

3

3

3

0

0

0

3

3

3

0

0

0

3

0

1

2 F k

2

2

0

2

2

0

2

2

0

0

2

0

0

2

0

0

1

4 F 1

4

4

4

0

4

4

0

4

0

4

0

0

4

0

0

0

1

4 F m

4

4

4

4

4

4

4

0

4

0

0

4

0

0

0

0

1

2 (Te

<rl <r'i

2 ^ e

<TI

^l 2

^ e

<rl <rl <rl <rl <

<

<rl 2

CTe

<^l •G^

8

EMS

- 2 4 96(^3

+ i6o(r^

+ 24oo | 2

+ 12aTrp

4 120(Tp

4 8oac,H . ^2

4 40(Tj).p

4 60Jj.jrp

^ ^2 4 60^Hp

4 300"^p

2 4 20(Tj)j^ip

^ 20<JDHP

2 4 100-j^ipp

^2 4 1 5 0 H T P

2 + 5^DHTP

55

TABLE 5

SUMMARY OF EXPERIMENTAL VARIABLES

Factor

Subject

Handwheel Diameter

Resistant Torque

Height

Hand Position

Level

1 2 3 4 5

rrJt

10" 14"

20 in. 40 in, 80 in. 120 in.

6" bel

-lb -lb -lb -lb

ow e 6" above e

60" left 60"right 120" 180*

Ibow Ibow

Code

1 2

5

1 2 3

1 2 3 4

1 2

1 2 3 4

Type

Random

Fixed

Fixed

Fixed

Fixed

"-.- ••!'

Identifi­cation

S

D

T

H

P

CHAPTER IV

ANALYSIS OP RESULTS

Analysis of Force Traces

As mentioned previously, there are three primary

techniques popularly used in analyzing recorded force

platform output. Two of the techniques require measure­

ment of the area under the force-time curve which has

stimulated some debate over just what this area repre­

sents and the validity of using it as an index of work,

energy expended, or fatigue. One may use the third method,

which does not require measurement of this area, if the

task involved is not complex and is of short duration.

In this method the maximum height of the trace is used

as an index of the forces exerted by the operator in per­

forming the task.

Since the previously mentioned requirements were

met by the task considered here, and since it was felt

that force would be a simple yet meaningful measure, a

variation of the third method was selected for use in

this experiment. The resultant of the average maximum

forces over five cycles (partial revolutions) produced

for the frontal, lateral, and vertical axes was used as

a single measure of operator performance. This resultant

average maximum force for a particular trial was computed

56

57

using values obtained from the force trace for that trial

produced by the Dynograph Recorder. A typical recording

is shown in Figure 12.

Determination of the resultant force for each

trial Involved five steps: (a) noting the maximum distance

either above or below the zero or static weight line for

each cycle for the three axes (see Figure 12), (b) adding

the five values to obtain the total maximum deflection

over five cycles for each axis, (c) converting these totals

to total maximum force for each axis by multiplying by a

constant obtained from the force platform calibration re­

cord (see Appendix A), (d) dividing the resulting totals

by five to obtain the average maximum force for each axis,

and (e) computing the resultant force by taking the square

root of the sum of squares of the average maximum force of

each axis.

The data thus obtained were punched on cards and fed

into a digital computer which performed an analysis of

variance.

Analysis of Variance

The analysis of variance was performed using com­

puter program BMD02V--Analysis of Variance for Factorial

Design—prepared by the Health Sciences Computing Facil­

ity, UCLA, and modified for the Texas Technological College

computer center. The resulting analysis of variance was

58

^^-^T^^tt^^ •Q

0 0

P - Frontal axis L - Lateral or transverse axis V " Vertical a cis

Pig, 12.--Typical Dynograph recording

59

then adjusted to reflect the desired design for this

experiment--a factorial experiment in a randomized block

design. The analysis of variance for this design appears

in Table 6, Only the main effects and interactions that

were significant at the one and five per cent levels were

considered for analysis and discussion. Note that all

main effects and all first order interactions with the

exception of the height by hand position interaction were

found to be significant at these levels• Although the

variation between subjects was found to be significant

(as might be expected), this effect will not be discussed

in great detail as it is not of major concern here and adds

little of significant value to the discussion. A graph of

this main effect is presented in Appendix B, It shows that

subject number four had a considerably higher mean result­

ant bodily reaction force than the other subjects. Subject

4 was the eldest of the group and had consistently larger

anthropometric measurements (see Table 2, Chapter II),

The subject interactions showed relatively little differ­

ence in subject reactions to changes in design parameters

and Indicated the same basic trends for each subject. Con­

sequently, subject interactions will not be discussed here.

The means of the remaining main effects and interactions,

however, were plotted to facilitate subsequent analysis

and will be discussed in detail.

TABLE 6

ANALYSIS OF VARIANCE

60

Source of

Variation

Subjects Handwheel diameter Torque

Height Hand position Handwheel dla, X torque Handwheel dla, X height

Handwheel dia, X hand position Torque x height

Torque x hand position

Height X hand position

Handwheel dia, x torque x height

Handwheel dla, x torque x hand pos.

Handwheel dia, x height X hand pos. Torque x height x hand position Handwheel dia. x torque x height x hand position

Residual (error)

Total

Degrees of

Freedom

4 2 3 1 3

6

2

6

3

9

3

6

18

6

9

18

380

479

^ — . - • . —

Sum of Squares

3806.09 7793.85 21432.42

1006.98 21878.01

1672.14

295.96

449.75

329.90

4592,92

134.00

171.70

602.81

163.90

208,15

225023

8241.44

73005.25

Mean Square

951.52 3896.92 7144,14

1006.98 7292,67

278.69

147.98

74,96

109.97

510.32

44.67

28.62

33.48

27.32

23.13

12.51

21.69

F

43.87* 179.66* 329.37* 46.43* 336.22*

12.85*

6.82*

3.46*

5.07*

25.53*

2.06

1.32

1.54

1.26

1.07

0.58

*Slgnificant at .01 level

61

Analysis of Graphic Data

In the remainder of this chapter we shall analyze

and discuss the significant effects graphically presented,

Handwheel Diameter Effect (Figure 13)

As is apparent from the graph, increasing the size

of the handwheel control decreased the amount of bodily

force produced in performing the task. The smallest hand-

wheel required the most force and the largest handwheel

required the least. This result was expected and is not

particularly surprising. Had this result not been obtain­

ed, however, the validity of the experiment would have

been suspect.

A Duncan Multiple Range Test as described in Hicks

(17) was performed on the three means. The following sym­

bols will be used in performing the test:

MSg = Error mean square (from Table 6)

k = number of means

p — '^9 r> > ' • • i k

N = number of observations in the mean

n = degrees of freedom of error term (Table 6)

s = standard error of a mean = YMS^/N

R = significant range from Appendix Table

E of Hicks (17)

LSR = least significant range = Rs

62

5 0 "

40 -

ra -CJ C ps o PL,

0 o u o p

30 "

20 •-

10 -•

0 1 1 h-7 10 14

Handwheel Diameter (in^)

Fig. 13.—Relationship between the resultant bodily reaction force and handwheel size

63

The procedure used in performing the test is shown

below in some detail in this instance, but hereafter only

the final results will be presented. For handwheel diam­

eters of 7, 10, and l4 inches the means were 30,00, 23.50,

and 20^51 respectively.

s = MS E 21.69 = 0^368 \ N 1 l6o

P = _2 3_

R = 3.64 3.80 (at one per cent level)

LSR = 1.34 1,40

Arranging the means in ascending order:

3 2 1 20.51 23.50 30.60

Range 1 - 3 - 9 . 4 9

Range 1 - 2 rr 6.50

Range 2 - 3 = 2.99

Comparing observed ranges between means with the

least significant ranges (LSR), we find:

1 versus 3 - 9.49 > 1.40 1 versus 2 = 6.50 > 1.34 2 versus 3 = 2^99 > 1.34

Hence, all means are significantly different•

They may be depicted as follows:

Handwheel diameter 7" 10" l4"

Mean 30.00 23.50 20^51

Hereafter, any means not underscored by the same line are

significantly different and any means underscored by the

64

same line are not significantly different.

It can be concluded from the previous discussion

that a significantly different amount of force is pro­

duced in the operation of each size handwheel and that

the 14 inch diameter handwheel is the optimum size of

those sizes tested since the bodily reaction force was

lowest for this size. The low bodily reaction force for

this diameter reflects the low applied force required.

Increasing the diameter increases the moment arm of the

applied force, thereby decreasing the applied force since

this moment arm and applied force constitute a torque in

opposition to any resistant or braking torque.

Note, however, that there is not as much advantage

gained in going from the 10 inch to the l4 inch diameter

as there is in going from the 7 inch to the 10 inch diam­

eter. This reduction in rate of gain is due to that dis­

advantageous aspect of increasing handwneel size previously

mentioned. As the handwheel diameter is Increased from 10

to l4 inches, the center of gravity of the body is displac­

ed enough to create a slightly unbalanced condition. This

slight imbalance brought about by the increased moment arm

causes the body to apply counteracting reactive force in

order to maintain balance. This lessens, somewhat, the

advantage in going to larger size handwheels. As handwheel

diameters continue to increase in size, these reactive

forces Increase also and the advantage gained by increasing

65

the handwheel size becomes less and less.

Resistant Torque Effect (Figure l4)

The relationship between resistant torque and the

bodily reaction force produced in overcoming this resist­

ant torque is clear-cut. The greater the torque that

must be overcome, the greater the bodily reaction force

produced. Note the almost linear relationship depicted

in Figure 14. The results of the Duncan Multiple Range

Test are given below:

N :. 120

s - 0o424

P = 2 3 4

R ^ 3o64 3.80 3.90

LSR = 1.54 1.61 1.65

Torque (inch-pounds) 20 40 80 120

Mean l6,62 20.22 27.08 34,76 s s — i ^ — M H B i V S M ^ K i V ^ O — P ^ H M W O H W a * ^ . ! ' ^ O W ^ M B B ^ B B S J H K V ^ O a ^ V s ^ n E I H ^ a a i i * ^ ^

All means are significantly different at the one

per cent level. Thus, the optimum value of torque for

those tested is the smallest value (i,e,, 20 inch-pounds).

This is a reasonable result since the bodily reaction

force is a function of the applied force required to over­

come any particular value of resistant torque. As the

resistant torque increases in value, the applied force at

the handwheel rim must Increase correspondingly if the

resistant torque is to be overcome. Increases in the amount

66

5 0 -

4 0 -

ra Td

o OH

3 0 -

0 O PH O P

20..

10"-

0 -H 1 1 1 1 1— 20 40 60 80 100 120

Torque ( i n , - l b . )

Fig. l4.--Relationship between the resultant bodily reaction force and resistant torque

67

of applied force are reflected as increases in the com­

puted force platform output or, in other words, the re­

sultant bodily reaction force since an applied force

effects the transmittal of a related reactive force along

the main body members to the force platform. The final

reactive force sensed by the force platform will not

necessarily be of the same magnitude as the original ap­

plied force required to overcome a particular value of

resistant torque but there probably should be a linear

relationship between the two as implied by the graph in

Figure l4.

Relative Height Effect (Figure 15)

The graph in Figure I5 indicates that the bodily

force produced decreases as the relative height is in­

creased from six inches below the elbow to six inches

above the elbow. To determine if the force is signifi­

cantly different for these two relative heights, a Duncan

Multiple Range Test was performed. The results are given

below:

N = 240

s = 0.301

P = 2

R =3.64

LSR = 1,096

Height 6" below elbow 6" above elbow

Mean 26.84 22.50

68

5 0 - -

4o

ra Td PI ps o P.4

3 0 ••

0 o Pl o

20 ••

1 0 •"

1 6"

below elbow

— I 6"

above elbow

Relative Height

Fig. 15,--Relationship between the resultant bodily reaction force and relative height

69

Thus, produced force at the two relative heights

are significantly different at the one per cent level.

This result in conjunction with the result of the analysis

of the graph In Figure 15 indicates that a relative height

of six inches above the elbow is the desired vertical hand-

wheel location for the task evaluated in this experiment.

However, this experimenter suspected that the basic con­

tributing factor here was not height, in itself, but the

angle of elbow flexion generated at each of the heights.

To check this hypothesis, the average anthropometric

measurements of the subject group (see Table 2, Chapter

II) was used to determine the average angle of elbow flex­

ion at each of the relative heights. From Figure l6 we

see that an angle of 75 degrees was generated for a rela­

tive height of six inches above the elbow. McCormick (24)

gives the results of a study by Provins and Salter which

showed that the optimum elbow angle is about 90 degrees

for a task requiring application of relatively large

amounts of force. The elbow angle at a relative height of

six Inches above the elbow is much nearer to this value

than that at the lower height.

Another important point is that at the lower height

a rotation action localized primarily at the elbow is gen­

erated, whereas at the greater height a rotation action of

the shoulder is generated. McCormick (24) also mentions

another study in which Provins was Involved which concluded

T Shoulder height

Elbow height

56,1'

43.5"

Angle A — 75 degrees

Angle B =: 132 degrees

70

Pig^ l6,--Angle of elbow flexion for the average subject

71

that "a rotation action of the shoulder has approximately

li times the force of a rotation action of the elbow, and

it has nearly 3 times as much staying power (the ability

to maintain a force)•" Thus, at the greater height more

force can be supplied as the demand for more applied

force increases before the aid of additional muscle groups

is required. Bringing additional muscle groups into play

is a more costly and less effective means of providing the

required force than where the force requirements can be

met by one muscle group without undue strain.

Finally, the fact that there is such a great differ­

ence in magnitude between the two angles probably accounts

for the significantly different amounts of reactive force

produced at the two heights.

Hand Position Effect (Figure 17)

As mentioned earlier in Chapter III, there were

four different sets of hand positions considered in this

experiment (see Figure ll). Two were rather unbalanced in

that there was an arc of only 60 degrees separating the

right and left hands. For one set, the right hand was

placed at top dead center of the handwheel and the left

hand to the left; for the other, the left hand was placed

at top dead center and the right hand to the right. These

sets are denoted as 60 degrees left (60°L) and 60 degrees

right (60°R) respectively In Figure 17• The other two

sets of hand positions are more balanced, with 120 and 180

72

50-

ra Td

o PL,

40-

30

ra 0 o Pl o &H

20"

10

0

60^L 60°R 120 180°

Hand Position

Fig. 17.--Relationship between the resultant bodily reaction force and hand position

73

degree arcs separating the right and left hands. The graph

in Figure 17 shows that as the hand positions are changed

from unbalanced to more balanced ones, the resultant

bodily reaction force becomes less. This seems to indi­

cate that the more balanced hand positions are less cost­

ly from a physiological standpoint. To determine if the

mean forces for these hand positions are significantly

different from each other, a Duncan Multiple Range Test

was run.

N = 120

s = 0.424

P = 2 3 4

R = 3.64 3^80 3.90

LSR = 1,54 1.6l 1.65

Hand Position 60°L 6OOR 120° l80°

Mean 34.21 28.00 21,67 14.82

As shown above, the mean forces produced using each

of the hand positions are all significantly different from

one another at the one per cent level of significance.

Thus, the optimum set of hand positions of those considered

in this experiment seems to be the one where the hands are

180 degrees apart with the left hand placed midway along

an arc running counterclockwise from top to bottom dead

center of the handwheel rim and the right hand placed

directly opposite. The 60 degrees left position appears

to be the most costly and least desirable of the four.

74

An explanation for the preceding results can be

found upon examining the initial hand positions and what

happens as the right and left hands move through the 90

degree arc to their final positions. When the hands move

counterclockwise from an initial 60 degrees left position,

body weight is shifted almost entirely to the left side.

The right and left hands begin and end their movements

on the left side of a vertical line dividing the hand-

wheel into equal right and left portions. Consequently,

the forces applied by the right and left hands are applied

entirely to the left of the longitudinal axis of the sub­

ject's body. An extremely unbalanced downward force is

produced which contributes substantially to the resultant

bodily reaction force.

In moving from the 60 degree right position to the

final position the right hand traverses a 6o degree arc

on the right side of the handwheel and a 30 degree arc on

the left, while the left hand traverses its entire 90

degree arc on the left side of the wheel. Thus, the right

hand applies a primarily upward force while traversing 6o

degrees of its arc. However, a primarily downward force

is applied over an arc of 120 degrees. Here again an

unbalanced force is applied, though not as unbalanced as

in the 60 degrees left position where the total l80 degree

arc traversed by both hands is located on one side of the

handwheel.

75

The 120 degrees position is a relatively balanced

position so there is a fairly small shift in body weight

as the right and left hands move through the first 60

degrees of their arcs. However, as the right hand moves

counterclockwise through the final 30 degrees of its arc,

it passes top dead center of the handwheel and joins the

left hand on the left side of the handwheel. Thus, an

unbalanced force is applied during a total arc of 60 de­

grees (30 for each hand) as compared to 180 and 120 de­

grees for the 60 degrees left and 60 degrees right posi­

tions, respectively.

The 180 degrees position is a perfectly balanced

one. The right and left hands are diametrically opposite.

The right hand traverses its 90 degree arc entirely on the

right side and the left hand traverses its 90 degree arc

entirely on the left side. Consequently, there is no large

unbalanced force generated which might substantially in­

crease the normal reactive force since the right and left

hands remain exactly opposite one another throughout their

arcs, and the forces they apply are in opposite directions

and of approximately the same magnitude,

Handwheel Diameter and Resistant Torque Interaction (Figure I8)

Figure 18 indicates that there is not much difference

in the bodily reaction force for the three handwheel sizes

at the lowest torque value. However, as the resistant

76

ra -d P: P3 O pL,

0 O u o p

50

40 -

30

O 7" diameter A 10" diameter D 14" diameter

20 ..

10 .-

0 + '+-20 40 80

Torque (in.-lb)

120

Pig, l8,--Relationship between resultant bodily reaction force and resistant torque for various handwheel sizes

77

torque increases, the difference in the forces becomes

more pronounced. The effect of increasing torque seems

to have the greatest impact on the smallest handwheel

diameter. Note that the largest handwheel actually had

a higher value of reactive force than the 10 inch diameter

wheel at the lowest torque, but that the largest handwheel

became more advantageous as the torque increased. The

advantage shifts to to the l4 inch handwheel somewhere

between 20 and 40 inch-pounds of torque. This is the

divergence point that Davis (6) found in his study. He

says that this point usually lies around the 40 inch-

pound torque. Although this study is of a considerably

different nature than Davis', it seems to offer addition­

al evidence that such divergence points exist and should

be considered in the design of handwheel controls.

These divergence points are due to the two primary

effects of increasing the handwheel diameter. The major

effect noted when the diameter becomes larger is a reduc­

tion in the required applied force and, consequently, the

reactive force. The second effect acts to increase the

reactive force. It is usually relatively minor but be­

comes more important as handwheel diameters get larger

and larger^ This effect was discussed in Chapter III

where it was noted that increasing the diameter and, in

effect, the moment arm of the applied force should increase

the bodily reaction forces to some extent for even those

78

handwheels that are not inordinately large. These two

antagonistic effects, then, interact to determine the

bodily reaction force for a particular size handwheel.

At low torque values and where larger handwheels are used,

the normally minor adverse effect of increasing handwheel

diameter may become the dominant factor in determining

the reactive force• As resistant torque increases, how­

ever, this adverse effect becomes less and less important.

Figure 18 shows that the most advantageous handwheel diam­

eter at 120 inch-pounds of torque is a diameter of l4

inches•

Handwheel Diameter and Relative

Height Interaction (Figure 19)

The graph in Figure 19 indicates that Increasing the

relative height does not have as pronounced an effect on

the larger handwheels as on the 7 inch diameter handwheels

This is reasonable when one considers that a rotation action

of the elbow is produced in turning the small handwheel at

the lower height, which is less effective in the application

of required force. As the relative height increases, the

locus of rotation moves upward toward the shoulder. For

the larger handwheels, a rotation action of the elbow and

of the shoulder is generated at the lower relative height,

although the elbow action is the primary action. As the

relative height becomes greater, the relative difference

In rotation action is much greater for the 7 incn diameter

79

ra Td Pl ps o pu,

0 O u o p

50 -

40 -

O 7" diameter A 10" diameter D i4" diameter

30 -•

20 "

10 ••

0 •+• •f

6" below elbow

6" above elbow

Relative Height

Pig, 19•--Relationship between resultant bodily reaction force and relative height for various handwheel sizes

80

wheel.

Figure 19 also shows that differences in reactive

force among the three handwheel sizes are not as great

at a relative height of six inches above the elbow. This

is probably due to the fact that the primary rotation

action is now centered about the shoulder for all three

handwheels, and the difference in elbow angles is com­

paratively small.

The most desirable combination appears to be a

handwheel diameter of 14 Inches at a relative height of

six Inches above the elbow.

Handwheel Diameter and Hand

Position Interaction "" (Figure 20)

The deleterious effect of the small handwheel is

quite apparent in Figure 20, The bodily reaction force is

substantially greater than for handwheel diameters of 10

and l4 inches. There is a marked difference in the reac­

tive forces of all handwheels for the 6o degrees left

position. This is not surprising as this is the most un­

balanced and least desirable hand position of the four and

the large unbalanced forces generated for this hand posi­

tion greatly accentuate the normal differences in reactive

force that are a function of handwheel diameter. At the

other three hand positions there is not a great deal of

difference in the bodily reaction forces for the 10 and 14

inch wheels, although the l4 inch handwheel appears to

81

ra Td c ps o

(XH

.\ 0 O PH

o p

50 •

40-

30 "

20 .

10 .

0 1—

0 7" diameter A 10" diameter D 14" diameter

1 1 —1

6 0 O L 6 0 O R 1200 i8oo

Hand Position

Fig. 20*--Relationship between resultant bodily reaction force and hand position for var­ious handwheel sizes

82

consistently show somewhat lower bodily reaction forces.

For these larger diameters and relatively more balanced

positions, the handwheel effect has the greater impact

and acts to attenuate the hand position effect to some

degree,

At hand positions of 60 degrees right, 120 degrees,

and 180 degrees the final displacement of the right hand

to the left of the vertical diameter of the 10 inch hand-

wheel is 2,5, 2.5, and 0 inches, respectively. For the

l4 inch diameter, these displacements of the hand that

determines to a large extent the relative imbalance of

the applied force is very slight for these hand positions.

However, for the 60 degrees left position the right hand

traverses its full 90 degree arc on the left side of the

vertical handwheel diameter. In other words, the final

displacement of the right hand is equal to the radius of

the handwheel for this hand position. The difference In

displacement for the two handwheels now becomes twice what

it was at the 60 degrees right position. The greater dis­

placement for the larger handwheel creates a larger moment

arm which probably decreases the magnitude of the unbalanc­

ed force applied by the right hand enough to more than off

set a slight increase in imbalance also generated by this

larger moment arm.

The general trend of the curves in Figure 20 indi­

cates the desirability of using balanced hand positions.

83

The most effective combination appears to be a hand posi­

tion of 180 degrees and a handwheel diameter of either

10 or 14 inches.

Resistant Torque and Relative

Height Interaction" (Figure 21)

The trend of the curves in Figure 21 reflects the

general torque effect of increasing reactive forces as re­

sistant torque increases. Note, however, that there does

not seem to be much difference in the forces for the two

relative heights at 20 and 40 inch-pounds of resistant tor­

que, but that the difference becomes more pronounced as

torque continues to increase. This is not an unreasonable

result since one would expect the advantages gained by using

the "better" height to become increasingly more apparent

as greater applied forces are exerted to overcome compara­

tively large amounts of resistant torque. Note also that

not only is there little difference in reactive forces be­

tween the two heights at the two lower torque values, but

that there is little difference between forces generated

at the 20 and 40 inch-pound torque values for each height.

But then, the increase in torque is only 20 inch-pounds as

compared to increments of 40 inch-pounds as torque is further

increased.

The analysis of Figure 21 seems to indicate that in­

creasing the relative height does not have a particularly

great impact on the bodily reaction forces produced at low

85

values of resistant torque, but that the impact is consid­

erably greater at higher torque values. At a resistant

torque value of 120 inch-pounds, for example, the advan­

tage gained by Increasing the relative height is quite

substantial.

Resistant Torque and Hand

Position Interaction (Figure 22)

Figure 22 shows that lower forces resulted from

the more balanced hand positions. As torque increases

above 40 inch-pounds, the differences in reactive force

among the various hand positions become appreciably more

distinct. The slopes of the top three curves indicate that

the resultant forces for these hand positions are markedly

affected as resistant torque is increased beyond 40 inch-

pounds • At the 20 inch-pound torque the difference be­

tween the two balanced and two unbalanced hand positions

is quite apparent, however, there is not much difference

between the 60 degrees right and 60 degrees left positions

nor between the 120 and l80 degrees positions at this torque value.

Note the relatively flat slope of the bottom curve

in Figure 22, This would seem to indicate that resistant

torque had considerably less effect when the most balanced

hand position was used. Such a hand position seems to

counteract somewhat the adverse effect of increased re­

sistant torque. The most logical reason for this is

86

50

40 -

ra Td P: P5

o pL. 0 O PH

o p

30 •-

20 "

10 ..

0

O A D d

60 degrees left 60 degrees right 120 dei

20 40 80 120

Torque (in,-lb)

Fig. 22,--Relationship between resultant bodily reaction force and resistant torque for various hand positions

87

probably the fact that the force platform is so much more

sensitive to unbalanced positions than balanced ones.

Finally, recall that after each subject had complet­

ed the complete series of trials, he was asked which combi­

nation (s) of variables he preferred. The majority of sub­

jects selected the l4 inch handwheel with 20 inch-pounds

of resistant torque, a height of six inches above the elbow,

and the l80 degree hand position• This closely agrees with

the results of the analysis presented in the previous sec­

tions of this chapter, and seems to bear out the hypothe­

sis that the human being instinctively seeks to perform,

a task requiring considerable effort in such a manner as

to minimize the amount of energy expended or force exerted.

In other words, the human being, when given a choice, will

usually select optimum or near optimum values of variables

affecting his performance of a task requiring major expen­

diture of energyo Konz and Day (21) found this to be true

in their study and Ellis (8) reported similar results in

an experiment evaluating work-surface height. At the con­

clusion of the experimental session, Ellis requested each

subject to adjust the work surface to the height at which

he (the subject) believed it would be most comfortable to

work^ An average preferred work-surface height of 4l,3

inches was obtained from this procedure, which was in close

agreement with the optimal value experimentally determined

to be 42,0 Inches (8).

CHAPTER V

SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS

The material in this chapter is presented in three

sections. The initial section gives a brief summary of

the entire study. The next section presents the conclu­

sions drawn from this research along with some personal

observations of the author. The final section points out

some areas worthy of further research.

Summary

This experiment was performed to determine the ef­

fects of handwheel size, resistant torque, control height,

and hand position on a simple, dynamic task involving the

operation of handwheel controls.

Three different sizes of handwheel controls and

four values of resistant torque were selected for consid­

eration. Also evaluated were two relative control heights

and four different hand positions.

A homogeneous sample of five subjects was selected.

Each of the subjects was required to rotate a handwheel

counterclockwise against resistant torque while standing

on a force platform, and were required to perform this

task for all combinations of variables.

The experimental design chosen was a factorial

88

89

experiment in a randomized block design, where subjects

were treated as blocks. Using this design, an analysis

of variance was performed on the data obtained from the

recorded force platform outputs in order to determine if

the variables had a significant effect on task performance.

A graphical analysis was then prepared to ascertain the

exact nature of these ef.fects.

Conclusions

Recall that several hypotheses were made prior to

this investigation (see Chapter III), Recorded evidence

with subsequent engineering and statistical calculations

sustained them all. These hypotheses were made based upon

logic and proven techniques of engineering analysis. Al­

though it is doubtful whether some of the more subtle re­

lationships between design parameters could have been

predicted in this fashion, it does indicate that a thorough

engineering analysis can be extremely valuable in predict­

ing the more fundamental effects of these parameters on

human performance. This is especially true where experi­

mental results may prove too restrictive, or where experi­

mentation might be considered too costly from the stand-

poing of time, labor, and money.

In many cases, however, the engineer has become so

infatuated with experimentation that he has forgotten the

basic tools of his profession. Care was taken to utilize

90

engineering principles and analysis techniques wherever

applicable and appropriate in conjunction with this ex­

periment as an aid in promoting sound, effective research.

Nonetheless, upon examining the results of this experi­

ment, it soon became apparent that the effects of some of

the variables could have been predicted and/or justified

using the basic principles of motion economy. Especially

applicable were those principles advocating symmetrical,

balanced motions of the body members. Although these prin­

ciples come from the very core of a specific branch of

industrial engineering, they apparently are generally ap­

plicable to many other related areas. This seems to pro­

vide additional evidence that research involving experimen­

tation which also incorporates effective techniques of en­

gineering analysis and the application of proven engineer­

ing principles usually produces results that are more gen­

erally applicable, more conclusive, and more easily sub­

stantiated than results produced by experimentation alone.

The conclusions drawn from this research were based

mainly upon evidence shown graphically and statistically

to be of practical significance. These conclusions are

additionally supported by basic scientific and engineering

principles as well as results of previous research. They

are :

1^ Resultant bodily reaction force can be used as

an effective index of physiological cost to the

91

operator in the performance of a simple, dynamic

task of the nature of the one evaluated in this

experiments The results of this study also of­

fer additional evidence that the force platform,

when properly used, can be a valuable tool in

the design and evaluation of hand controls.

2. Larger handwheels are more advantageous to the

operator at resistant torque values above 40

inch-pounds. Handwheel diameter should be in­

creased as resistant torque increases above

this value in order to decrease the applied

force requirements. Care should be taken,

however, to insure that diameters do not be­

come so large that they become unwieldy, and

more costly from a physiological standpoint,

3. As the resistant torque that must be overcome

decreases, the applied force requirements and

the physiological cost to the operator become

less until a torque value of 40 inch-pounds is

reached. As torque values decrease below this

value, the resultant bodily reaction force may

or may not decrease depending upon the handwheel

size. When designing a handwheel control to be

used in a task such as that evaluated here, the

combined effects of the proposed handwheel size

and resistant torque should be taken into

92

consideration. The two parameters should be

evaluated as a combination and not separately.

4. Less physiological cost is generated when hand-

wheel controls are operated such that an elbow

angle of near 90 degrees is maintained and,

where this angle cannot be maintained, angles

less than this value are generally preferable

to angles exceeding 90 degrees.

5. Balanced hand positions are more advantageous

than positions where both hands are on the same

side of the handwheels This is especially true

at higher values of resistant torque where a

hand position of l8o degrees proved to be un­

questionably the best of those evaluated•

This experiment was, of necessity, somewhat restric­

tive in nature• Extreme caution should always be exercised

in attempting to generalize the results of such an experi­

ment. Nevertheless, it is felt that the findings of this

study are generally applicable to tasks similar to the one

considered in this experiment, and provide valuable Infor­

mation that can be extremely useful in the design of hand-

wheel controls for such tasks.

Recommendations for Further Research

The experimental evidence and conclusions obtained

from this study indicate areas in which further research

93

is likely to produce more specific and useful results.

The following items deserve the attention of further re­

search:

1. Smaller increments of resistant torque at the

lower values should be evaluated experimentally

to pinpoint more accurately the divergence

point above which the advantage definitely

lies with the larger handwheels.

2. An effort should be made to experimentally

determine the effect of very large handwheels.

3. In addition to the preceding direct variations

of this experiment, a number of other possible

applications of the force platform are indicated.

Use of the force platform could probably indicate

the onset of physical fatigure by indicating lack

of coordination and decreased use of an ability

in worker performance. It may also enable the

detection of differences in work methods more

precisely and easily than other physiological

measuring methods. Finally, the force platform

may be found useful as a means of determining

industrial efficiency and classifying jobs by

physiological cost.

LIST OF REFERENCES

1. Adams, S. I965. Determination of human mechanical energy and work output from independent measures of force and motion. Unpublished PH.D. dissertation, Arizona State University,

2. Barany, J. 1963. The nature of individual differ­ences in bodily forces exerted during a motor task as measured by a force platform. Journal of Industrial Engineerings l4: 332-41.

3. Barnes, R, - 1963. Motion and time Study. New York: John Wiley.

4. Corlett, E. 196I. The accuracy of setting of machine tools by means of handwheels and dials. Ergonomics. 4(l): 53-62.

5. Damon, A., Stoudt, H,, and McFarland, R. 1966. The human body in equipment design. Cambridge, Mass.: Harvard University Press.

6. Davis, L. 1949. Human factors in design of manual machine controls. Mechanical Engineering. 71:811-16.

7. Edel, Jr., D., ed. I967. Introduction to creative design. Englewood Cliffs, N. J.: Prentice-Hall, Inc.

8. Ellis, D. 1951. Speed of manipulative performance as a function of work-surface height. Journal of Applied Psychology, 35^ 289-96.

9. Floyd, W,, and Welford, A., ed. 1954• Human factors in equipment design. London: H, R, Lewis and Company,

10. Greene, J,, and Morris, W. 1958, The force platform: an industrial engineering tool^ Journal of Industrial Engineering. 9° 128-132.

11. Greene, J., Morris, W., and Wiebers, J. 1959. Nine methods for measuring physiological cost of work. Journal of Industrial Engineerings 10: i8oi^5ir:; ^ _ — -

94

95

12. Greene, J., and Morris, W. 1959. The design of a force platform for work measurement. Journal of Industrial Engineering. 10: 312-17, — ^ --^

13. Greene, J. 1957. Design and initial evaluation of a force platform for measuring human work. Unpublished PH. D. dissertation. State University of Iowa, (cited in Hudson, 1962).

14. Handbook of htmian engineering data. 1952• Medford, Mass.: Tufts University.

15. Helson, H. 1949. Design of equipment and optimal human performance. Am. Journal of Psycho­logy. 62: 473-97. " """^

16. Hertzberg, H. I96O. Some contributions of applied physical anthropology to human engineering. USAF, WADD, Technical Report 6O-I9.

17. Hicks, C. 1964, Fundamental concepts in the design of experiments. New York: Holt, Rinehart, and Winston.

18. Hudson, W. 1962. The study of the dynamic character­istics of a reaction free' apparatus for measur' ing physiological cost of dynamic manual work^ Journal of Industrial Engineering, 13: 15-I8.

19. Ingenohl, I, 1959. Measuring physical effort. Journal of Industrial Engineering, 10: 99-114,

20. Karnasiewicz, E. 1968. An electromyographic study of a rotary task. Unpublished thesis, Texas Technological College,

21. Konz, S,, and Day, R. 1966. Design of controls using force as a criterion, H'oman Factors, 8: 121-27.

22. Lauru, L., and Brouha, L. 1957- Physiological study of motions. Advanced Management. 22: 17-24,

23. Lehmann, G, 1958. Physiological basis of tractor design. Ergonomics. 1: 197-206.

24• McCormick, E. 1964. Human factors engineering. New York: McGraw-Hill.

96

25. Morgan, C, Chapanis, A,, Cook, J., and Lund, M, 1963. Human engineering guide to equipment design. New York: McGraw-Hill,

26. Slnaiko, H., ed, 1961. Selected papers on human factors in the design and use of control systems. New York: Dover Publications, Inc.

27. Taylor, E, 1959- The interim report on engineering design, Cambridge, Mass,: Massachusetts Institute of Technology.

28. Whitney, R, 1958. The strength of the lifting ac­tion in man. Ergonomics, 1(2): 101-28

29. Woodson, W, 1954. Human engineering guide for equipment designers. University of " California Press.

APPENDIX

A. Force Platform Calibration Record

B. Graph of Subject Main Effect

97

98

APPENDIX A: FORCE PLATFORM CALIBRATION RECORD

Applied Force

(lb)

0

5

10

15

20

25

30

Pen Deflection in Millimeters

Frontal Axis

0.0

4,0

8.0

12.0

l6.0

20.0

24.0

Lateral Axis

0.0

4.0

8.0

12.0

16.0

20,0

24.0 •

Vertical Axis

0.0

2.0

4.0

6.0

8.0

10.0

12.0

ra Td

o pL.

0 o u o

99

APPENDIX B: GRAPH OP SUBJECT MAIN EFFECT

50

40-

30--

20--

10 .-

O O

o o

0 3

Subject Number